Reliable Change Index
Test scores will very from one administration to another due to imperfect reliability. In order to test if the change in scores is greater than might would be expected from random variation alone, many researcher's advocate the use of the Reliable Change Index Statistic. If the change score exceeds the Reliable Change Index (which is calculated based on the reliability of the measure), then the patient can be said to be "significantly improved", meaning that the observed change would be expected by chance alone at a probability of less than 5%.
The formula for the Reliable Change Index is:
square root (2 X (
standard error of measurement)2).
I have attached an Excel utility to calculate the Reliable Change Index and clinical cutoff score for any outcome measure.
The largest problem with a simplistic use of the RCI statistic is that it does not take into account regression artifacts: patients reporting more severe symptoms are much more likely to show significant change on the outcome questionnaires that those reporting mild symptoms. See
RegressionToMean
For more information, see
Dr. Lee Becker's discussion of
reliable change.
Primary topics:
You may also view all topics in alphabetical order in
WebTopicList.
--
JebBrown - 15 April 2007